How to Find a Parenting Effect

Like religion and politics, parenting can be an emotionally charged topic. I argued previously that parenting did not represent a monolithic predictor of child development.^1,2,3,4 More precisely, I stated that even if it was, most research on the subject would never allow you to know it because of a problem replete in the social sciences: genetic confounding. The larger intent of my previous essay, in fact, was to address the perils of correlational research and encourage you to think carefully the next time you saw a headline proclaiming that X causes Y (e.g., bacon causes cancer). Parenting effects provided a suitable avenue for making that point given the proliferation of deeply confounded “parenting studies” which trade on unintelligible correlations between parenting styles and child outcomes. I want to take a slightly different approach this time around. If parenting effects really existed, and you wanted to find them, where would you need to look?

You know already where not to look; a correlation between a parenting style and child behavior, for instance, simply does not provide enough information to conclude that the former causes the latter. Many researchers, thankfully, understand this point on some level.¹ In an ironic twist, actually, it has become fashionable of late to embrace genetic effects on personality, at least to some degree. It seems every social scientist can rattle off the following mantra: “human development is a complex interplay between nature and nurture” or some version of that.¹

Practically no one these days will fight you on the idea that genes and environments interact (GxE), and it is in these interactions, perhaps, that we can discover if parenting truly does impact children.¹ As you’ll see, though, this simple notion (gene-environment interaction), belies some complexities that can complicate life quickly. If we’re not careful, they will beguile our ability to truly find a parenting effect, assuming one exists to be found.

For now, at least, we’re all happily on board the train headed toward vindicating the existence of those elusive parenting effects. No sooner than we start building some steam, though, do we find that an absent-minded farmer has let a herd of cattle escape and they’re now grazing on our track bound for knowledge. That wandering herd in a methodological sense is something called gene-environment correlation (rGE). Genes and environments are not tossed together randomly.⁴ Einstein once famously quipped that: “God doesn’t play dice” [to loosely paraphrase the great genius]. The man with the enviable head of hair was commenting on the nature of universe, of course, not parenting and child development. Had he been talking about parenting, I like to think that he might have added: “child rearing isn’t a roulette wheel.” Bright parents often have bright children (to use a well worn example). These bright individuals—owing to their own tastes and preferences—are likely to rear their children in homes full of technology, music, art, and culture.¹ In short, the child’s genetic propensity to be smart is reflected back in their direction from the environment. Behavioral geneticists refer to this as a passive gene-environment correlation (rGE) (parents provide both the genetics for high intelligence and the environment conducive to its successful manifestation).

Do Parents Make a Difference? A Public Debate in London

Sydney. London. Toronto.

QuilletteToby Young

But wait, there’s more. As any parent knows, children evoke reactions from their caregivers. Parenting is not generally a one-way street.¹ The reactions that children elicit are based in part on their genetically informed temperaments (known as evocative gene-environment correlation). Another reality for parents (one they may come to lament) is that as their children age they actively seek to construct for themselves an environment that aligns with their natural inclinations (active gene-environment correlation). The lamentation likely comes when the child begins “creating” environments that run counter to mom and dad’s preferences (parents with children who have begun dating, I’m sure, are nodding along with me). The reality of gene-environment correlation creates a methodological problem of some significance for researchers. If you’re going to tease apart that interaction between genes and environment (GxE), then the existence of gene-environment correlation (rGE) must also be dealt with. You must clear the cows from the track.

Now a true dissection of the problem can begin. If you read my first discussion on this topic, then you will likely recall that twin studies cut trait variation in the population into three categories: heritability (capturing the influence of genes on human differences), shared environment (capturing the influence of families and parents, as well as other factors that make siblings similar), and non-shared environment (capturing all of the unique experiences that we have across our lives, including random events, that make siblings dissimilar). The laws of behavior genetics³ suggest that genes and the non-shared environment play the largest role for most traits, but that’s a very general understanding (albeit an extremely well vindicated one). What about rGE and GxE? We know these complex processes are likely going on in the background. What is the consequence of not testing for them in developmental research? The behavioral genetics researcher Shaun Purcell has done the legwork for us (p.555)⁶ (note that A simply refers to heritability, C to the shared environment, and E to the non-shared environment):

Before considering the modelling of G × E it is worth reviewing the impact of G × E and r_GE on standard twin models, in terms of biased parameter estimates. In short, interaction between A and C acts like A; interaction between A and E acts like E. Correlation between A and C acts like C; correlation between A and E acts like A.

That first point is worth repeating, an interaction between a child’s genes and something in the shared environment (like parenting, perhaps) would show up as a genetic effect in the twin model.

Ah hah! Perhaps all those heritability estimates are masking interactions between genes and parental treatment? Yes, perhaps. But let’s retrace our steps for a minute. Remember that brilliant study by Polderman and colleagues⁷ (I mentioned in my previous essays). In reviewing fifty years worth of twin research, the general pattern suggested a relatively small impact of the shared environment—perhaps the genetic effects are sucking up all of the important parenting interactions (like a big Hoover vacuum or something). Don’t forget, though, that interactions between genes and the non-shared environment get lumped in as non-shared environmental effects. So we might be incorrectly classifying some effects as genetic (when they are not), but we might also be wrongly classifying some effects as environmental, as well. Then there is the issue of rGE and divvying up those effects. It seems complicated, but don’t let this bog us down. The long and short of it is that you need an approach capable of doing all of these different tasks (and in case you needed a reminder, classic correlational parenting studies fail on all fronts).

The psychologist Alexandra Burt, a few years back, conducted a review⁸ of behavioral genetic research in order to specifically examine the role of the shared environment on outcomes that cross my mind frequently as a criminologist: child and adolescent psychopathology. Burt was interested in conduct disorder and attention deficit disorders (as well as other traits), which tend to increase the likelihood that one will someday become a criminal (my own field of expertise). For those craving a more technical discussion of things, I highly recommend Burt’s paper. It’s both thoughtful and thorough. For those disdainful of journal articles, let me give you a brief sense of the findings. Cognizant of the problems we touched on earlier (undetected rGE in particular), Burt compared the results of twin studies on the topic of early life psychopathology with the findings of adoption studies (nearly 500 of them in total). Adoption studies are helpful because they sidestep certain issues that can artificially puff up a shared environmental effect (adopted children share zero genetic material with their parents, no genes are being passed along). The results suggested that the shared environment explained somewhere south of 30 percent of the variance for most (but not all) of the outcomes examined in the study. The effects were similar between both twin and adoption studies.

Heritability and why Parents (but not Parenting) Matter

Sydney. London. Toronto.

QuilletteBrian Boutwell and Razib Khan

Let’s ponder this for a moment. We’ve stumbled upon a significant shared environmental effect on psychological disorders in children, and it emerged in two types of behavioral genetic designs (twin and adoption studies). What to make of this? I think there are a few things you should consider. First, the effect of shared environments was lower in magnitude than the effects of genes or the non-shared environment, something predicted by the three laws of behavior genetics years ago. Second, we need to consider what it is about the “shared environment” that might actually matter. The term “shared environment” captures a fair amount of “stuff.” Deciding what precisely in the shared environment is the causal force is an important task⁸ and there is no guarantee that the causal agent (or agents) will involve parenting styles (though it could; see Burt’s work for a more thorough discussion).

The psychologist Robert Krueger and his team provided another example⁹ of how to most effectively probe the issue of whether, and to what extent, genetic factors interlace with parenting styles (in the form of both rGE and GxE). In this case, the researchers were interested in whether genetic effects on personality traits (like positive and negative emotionality) were altered based on the respondent’s perceived relationship with their parents. The results seemed to suggest that genetic influences on personality styles either increased or decreased based in part on the nature of the parent-offspring relationship. For instance, when perceived conflict between parents and respondents was high, genes had less of an impact on negative emotionality compared to when conflict was lower (genetic influences mattered more in that case); an interesting result to be sure. But what did the research team make of their findings, as a whole (p.1515)?

Clearly, these findings of shared environmental influence require replication before we make too much of them, but they do suggest a new vantage point for trying to understand how environments affect personality, made possible by the modeling advances portrayed in Figure 2 (the biometrical moderator model). The literature to date is not “wrong” in documenting the trivial impact of shared environments on personality in general. For most adolescents, in families with normative levels of conflict, personality does result from genetic factors (A) and environments that make people different from their family members (E; see table 2). Nevertheless, our understanding can be enhanced by examining adolescent personality in unusual family circumstances—in our case, unusually conflicted or unconflicted relationships with parents—where the impact of the family-level, shared environment (C) on personality can be seen.

I concur with the authors, but this only serves to buttress a point I made initially. In most cases (that is, on average), most of the differences in the population for personality traits are attributable to genetic differences, not parenting. As the literature continues to build, though, we will continue to refine our understanding in this regard.

I’d like you to shift your focus a bit now in order to revisit the topic of identical twins. In my last essay I focused on their stunning similarity, but in this case I want you to consider their differences, as they might provide us another avenue to sniff out those parenting effects.¹ The psychologist Judith Rich Harris made the point that gene-environment interactions with the shared environment, while interesting, cannot really explain differences that emerge for identical twins raised together.¹ Why not? The answer is that they don’t differ from one another genetically (not in any appreciable manner). If there was an interaction happening, it should impact both twins thereby making them more similar (interactions with the non-shared environment, of course, could create differences, but that’s a different topic). Harris devotes an entire book to dealing with this and other issues related to human individuality (No Two Alike), and I strongly urge you to read it. Nonetheless, if we seek to explain why identical twins turn out differently, then we must probe the issue in a slightly different way. GxEs will not unlock all the mysteries after all (see chapter 3 in Harris for a superb discussion on the misleading allure of GxE in developmental research).¹

Let’s imagine that mothers in the population tend to be more attached to some of their children than others (yes, I’m sure you love all of your children equally, just endure me on this point). Would that really mean much for development? Using a sample comprised only of identical twins, my colleague the biosocial criminologist Kevin Beaver investigated whether differences in types of parenting outcomes (maternal attachment, disengagement, involvement, and permissiveness) impacted delinquency in a national sample of American adolescents.¹⁰ He also tested whether these differences in parenting styles impacted levels of self-control (a primary predictor of antisocial and delinquent behavior). The design of the study (using only identical twins) is important because by its very nature it overcomes so many of the problems inherent with classical parenting studies. What were his findings? Differences in the parental relationships of the participants had virtually no consistent impact on the behavior of the participants.

Along similar lines, I would point your attention to a study led by the psychologist Avshalom Caspi.¹¹ Also analyzing a sample of identical twins, Caspi and his team produced evidence that maternally expressed emotions did predict behavioral differences in their offspring. Two studies are not enough to sway our interpretation about anything too dramatically, and like all methods the monozygotic difference approach has limitations. But it does illustrate the broader point of this essay; real parenting effects are only to be found in research approaches like those described above, strategies that are designed to sidestep the pitfalls of traditional social science research (I would also encourage you to see the early seminal papers of Robert Plomin and colleagues on this topic^2,3 for more detail than I can provide).

Let’s switch gears entirely now and talk about the topic du jour in the social sciences: epigenetics.^12,13 Epigenetics, it would seem, has become a buzzword for those social scientists who are gene-phobic (I think some of them see the double helix in their nightmares).¹³ To understand why, you must first understand what the science of epigenetics entails. Though your genetic code is fixed, whether those genes are “turned on” or “off”, by necessity, changes over time. Think of your genome as an orchestra playing the music from Star Wars (may the force be with you). Different sections must be brought in at the appropriate time (no one wants to hear the tuba all the time, especially if the scene involves Darth Vader strolling on screen). Your epigenome (the chemical code residing above your DNA) is a conductor, of sorts, cuing up the woodwinds, strings, and brass at just the right moments [i.e., turning on the correct genes when necessary] (I take no credit for the orchestra metaphor. It came to me while writing this and after having just seen the new Star Wars movie. If you’re not a Star Wars fan, my apologies).

On Parenting and Parents

Sydney. London. Toronto.

QuilletteBrian Boutwell

Here’s the part, though, that really gets the social scientists buzzing. Epigenetic effects in humans can be influenced directly by environmental inputs. In other words, the environment of organisms, humans included, can impact how genes are expressed. See, the genome isn’t superior after all, they crow. This is clearly relevant to parenting, right? Epigenetics research has shown using rodents, for instance, that maternal care can alter gene expression in rat pups; a huge finding.¹² Could something like this be at work in humans? Sure it could. Environmental exposures could alter genetic expression, playing a key role perhaps in explaining the differences that emerge between siblings, even between identical twins.¹² Consider that by the time two identical twins have reached adulthood, they have had a wealth of experiences that are uniquely their own, separate from their genetic clone (i.e., their sibling). You can imagine the excitement and intrigue surrounding these findings (rightly so), and the embrace of epigenetics by the broader public has been swift. Epigenetic findings continue to be written about in the media fairly frequently and it has even penetrated the discourse on some well informed parenting blogs. Yet, as with any nascent science, we should take a deep breath and be patient. The psychologist Terrie Moffitt and the criminologist Amber Beckley made this point in a recent paper (p.124):¹²

Many social scientists embrace the new epigenetics research because it has been billed as evidence that environment trumps genes. There is much excitement about this approach, which promises to capture a biological signature left behind by environmental adversity. However, our reading, and that of many biologically oriented scientists, is that epigenetics has been wildly oversold, particularly in the media. Many of our expert epigenetics research colleagues are deeply embarrassed by the warm, uncritical response their work has attracted from the social sciences. A biologist attendee at a July 2014 Washington, DC workshop on the social and behavioral implications of epigenetics gasped “The biologists there were horrified at the thought…we really don’t understand the basic biology well enough yet to do this!” A social scientist attendee agreed, “After the meeting I got the feeling the popular media has sold us a false bill of goods.” Here, we briefly summarize several cautions but refer interested readers to more in-depth discussions and sobering views by real experts in epigenetics (Heijmans and Mill, 2012; Juengst et al., 2014; Mill and Heijmans, 2013).

Does this mean that parenting has zero effect on the expression of genes in children? No. What it does mean is that we simply do not have a mature knowledge base regarding epigenetic effects on development. More to the point, we are still accumulating evidence as to whether certain parenting strategies exert a causal influence on gene expression, which then exert a casual effect on child development that lasts for any appreciable amount of time. Keep in mind also that to study epigenetics in humans is to do so absent the luxury of the experimental designs that can be used with non-human animals. One doesn’t need to be a bioethicist to realize that experimentally switching children around in order to see how being with adopted mommy fiddles with gene expression would be just a bit unethical. This inability to do experiments carriers with it a serious methodological point that only further pulls the rug from beneath us. The team of Moffitt and Beckley note (p.125):¹²

To date, the most compelling reports of methylation effects have examined rodents (a key fact conveniently omitted in many popular media reports). In rodent models, all research participants are genetically identical, and everything in their environment is held constant across treatment and control groups except the experimental manipulation. This uniformity combined with experimental control over the severity of the treatment dose makes it relatively easy to detect effects in laboratory animals, compared with humans, who are characterized by staggering levels of both genetic and environmental diversity. As noted, ordinary observational studies cannot rule out selection effects, but the use of twins, especially longitudinal analyses in discordant monozygotic twins, can help to pin down and hold constant some of the complexity that will otherwise compromise translation of epigenetics research from rodent models to humans.

Be sure not to miss the final point made by Moffitt and Beckley. If you want to appeal to epigenetics in order to study parenting, you had better use the appropriate research designs. Otherwise, you are going to wrangle with the same or similar confounds that I berated in my first essay on this topic.

By now you should have a better idea of where to look when looking for parenting effects. My hope is that you can be a truly informed and discerning consumer of research on this topic moving forward, if you weren’t already. Lastly, I wanted to speak to a broader reaction that I observed in the wake of our first discussion of parenting (both in correspondence and in chatting with folks face to face): the issue of “parental responsibility.” To write an essay arguing that parents don’t exert a lasting influence on their child’s personality seems needlessly provocative. Think of the fallout if people believed the argument. Without vigilant parental guidance, might children spiral into moral depravity, becoming increasingly narcissistic, materialistic, and even atheistic (Zeus forbid!)? There is simply no good reason to suspect that this is true. Yet, don’t let the limitations of parental socialization lure you into a lazy conclusion, such as one embracing the idea that parents are absolved of their responsibilities in child rearing.¹

If a child has attention deficit hyperactivity disorder, for instance, a particular parenting style likely had nothing to do with it. The reality of the disorder, however, means that the parent is faced with a range of choices. They alone will have to decide how best to navigate the problem. They could simply do nothing, or they might embrace some bizarre notion that they can “discipline” the problem away, or that proper religious instruction will correct the child’s behavior. None of these options represents a particularly appealing approach if the desire is to help the child. Effective action on the part of the parents should by necessity involve something more nuanced (e.g., consulting a mental health professional), but the need for them to act remains unaltered. Parents still shoulder a huge burden; it just doesn’t include uploading a personality or moral sense into the heads of their children. Judith Rich Harris provided us the final word last time. This time, Merle Haggard sends us off to think more on the issue of parenting:

“And I turned twenty-one in prison doing life without parole, no one could steer me right but mama tried, mama tried. Mama tried to raise me better, but her pleading I denied. That leaves only me to blame, ‘cause mama tried.“

References

1. Harris, J. R. (2007). No two alike: Human nature and human individuality. New York, NY: Norton.
2. Plomin, R., & Daniels, D. (1987). Why are children in the same family so different from one another? Behavioral and Brain Sciences, 10, 1–60
3. Turkheimer, E. (2000). Three laws of behavior genetics and what they mean. Current Directions in Psychological Science, 9(5), 160-164.
4. Scarr, S., & McCartney, K. (1983). How people make their own environments: A theory of genotype→ environment effects. Child development, 424-435.
5. Plomin, R., Asbury, K., & Dunn, J. (2001). Why are children from the same family so different? Nonshared environment a decade later. Canadian Journal of Psychiatry, 46, 225–233.
6. Purcell, S. (2002). Variance components models for gene–environment interaction in twin analysis. Twin research, 5(06), 554-571.
7. Polderman, T. J., Benyamin, B., de Leeuw, C. A., Sullivan, P. F., van Bochoven, A., Visscher, P. M., & Posthuma, D. (2015). Meta-analysis of the heritability of human traits based on fifty years of twin studies. Nature genetics, 47, 702–70.
8. Burt, S. A. (2009). Rethinking environmental contributions to child and adolescent psychopathology: a meta-analysis of shared environmental influences. Psychological bulletin, 135(4), 608.
9. Krueger, R. F., South, S., Johnson, W., & Iacono, W. (2008). The Heritability of Personality Is Not Always 50%: Gene‐Environment Interactions and Correlations Between Personality and Parenting. Journal of personality, 76(6), 1485-1522.
10. Beaver, K. M. (2008). Nonshared environmental influences on adolescent delinquent involvement and adult criminal behavior*. Criminology, 46(2), 341-369.
11. Caspi, A., Moffitt, T. E., Morgan, J., Rutter, M., Taylor, A., Arseneault, L., Tully, L., Jacobs, C., Kim-Cohen, J., and Polo-Tomas, M. (2004). Maternal expressed emotion predicts children’s antisocial behavior problems: Using monozygotic-twin differences to identify environmental effects on behavioral development. Developmental Psychology, 40, 149-161.
12. Moffitt, Terrie E., and Amber Beckley. “Abandon twin research? Embrace epigenetic research? Premature advice for criminologists.” Criminology 53.1 (2015): 121-126.
13. Wright, J. P., Barnes, J. C., Boutwell, B. B., Schwartz, J. A., Connolly, E. J., Nedelec, J. L., & Beaver, K. M. (2015). Mathematical proof is not minutiae and irreducible complexity is not a theory: a final response to Burt and Simons and a call to criminologists. Criminology, 53(1), 113-120.